The Need for a General Theory of Discrimination: A Comment on Katharine T. Bartlett & Mitu Gulati, Discrimination by Customers

Professors Katharine T. Bartlett and Mitu Gulati raise a fundamental and mostly neglected point about antidiscrimination law.1 Ordinary moral intuitions do not entirely support a legal exemption for customers. If a person said that he preferred watching college hockey to college basketball because of the differing racial composition of the players, anyone accepting basic antidiscrimination norms would quickly label the person—a customer of sports entertainment—a racist. It would be considered strange, and no defense, for the person to add, “But I never discriminate when I sell things.” Our normative evaluations of racism are not so dependent on the particular role one momentarily occupies in an economy. Yet antidiscrimination law ignores this customer’s decision to discriminate on the basis of race, and more generally, any protected characteristic. This perhaps startling and certainly interesting omission has recently generated a flurry of academic interest.2

In the course of raising and exploring this issue, Bartlett and Gulati make a number of powerful points. In addressing whether the legality of customer discrimination is normatively justified, they are surely right to give attention to efficacy and privacy/autonomy. By efficacy, they refer to the possibility that the regulation of firms (who do most of the selling in the economy) achieves most of the benefits of the antidiscrimination agenda at lower costs than the direct regulation of individuals. By privacy and autonomy, they mean that there are sometimes particularly important values at stake in giving people associational freedom from regulation. Dating, for example, is a site of considerable discrimination, but the pursuit of romantic and sexual relationships is at the core of individual identity. Ordinary firms focus on making profit, which requires more fungible and less intimate associations. Bartlett and Gulati sensibly evaluate both positions, showing that, for a variety of reasons, neither one by itself fully justifies the law’s near total indifference to individual customer discrimination.

Despite much agreement, I offer three criticisms, the first two being rather brief. First, Bartlett and Gulati discuss customer discrimination, but they omit employee discrimination. Just as antidiscrimination law covers most sellers but excludes customers, it also covers employers but excludes employees (ironically, the sellers of labor). Employees violate no law by announcing that they will only work for firms or proprietorships owned or managed by people of the same race. If one is focused on the gap between common intuitions and law, then it also applies here: An individual who refused to work for a fast food franchise owned by members of another race would hardly be excused from a charge of racism on the grounds that she was acting merely as an employee. Also, if one wanted to integrate workplaces by race, employee discrimination poses as much of a barrier as employer discrimination. If no white, Asian, or Latino/a workers apply to work at a business with only black employees, then the workplace will remain segregated regardless of the employer’s willingness to hire other races.

Second, I wish Bartlett and Gulati had started their article with the simple observation that the law frequently and appropriately deviates from popular moral intuitions. Consider, for example, promise keeping. Contract law does not enforce every morally obligatory promise, but ignores promises made without consideration or detrimental reliance, which includes many gratuitous promises among family members.3 Similarly, the law of fraud, defamation, and perjury covers a lot of lies, but the law permits a lot of morally condemnable social lying, including those told for the purpose of sexual seduction.4

Why do law and moral intuitions come apart? At least when the divergence is normatively justified, it is usually for the reasons Bartlett and Gulati describe as efficacy and privacy. In some cases, immoral conduct causes little or no harm. Consequently, the administrative costs of using law to remedy the wrongs are not worth the meagre benefits. Also, much of the immoral conduct occurs in social settings where the intrusion of government regulation is thought to endanger personal privacy and autonomy, given the inevitability of errors in the application of law. We see this with lies and broken promises, many of which cause little or no harm, and frequently involve social relations, as lies told in intimate settings and promises made within a family. The law cannot save the administrative costs if it weighs costs and benefits on a case-by-case basis before deciding not to create liability where the harm is low. So the law makes a rough cut, creating the categories of breached promises and lies that usually do not cause much harm, permitting them despite their being perceived as wrong.

Bartlett and Gulati could have framed their explanation of nonliability for customer discrimination as a general application of the divergence of law and morality, the general point that ideal theory is always compromised by real-world concerns of administrative costs and enforcement errors. Even without this framing, however, Bartlett and Gulati do much of the right kind of analysis, looking for categories of individual customer discrimination that should trigger legal liability after weighing efficacy and privacy considerations. As they reasonably observe, “[s]ome types of customer discriminations are not as harmful as others, and some limits on customer discrimination are more intrusive than the alternatives.”5 We might add a point Saul Levmore makes, which is that the market might be better at eroding discriminatory norms in some domains than in others, and the need for law is less compelling where the market succeeds.6 Collectively, these points mean that antidiscrimination law, like other areas of law, does not simply track the lines of our moral intuitions.

And yet, despite significant agreement, I am left with one serious concern and my third and primary point: We cannot ultimately decide how we should evaluate the wrong of customer discrimination without returning to the basic theory for why discrimination is wrong, or as Deborah Hellman puts it, “when discrimination is wrong.”7 Perhaps this claim seems inconsistent with the point above, but it is not. Even though law does not perfectly track the common understanding of morality, it is logical to start with a general theory of the harm or wrong of discrimination, and then apply that theory to the particular context of customer (or employee) discrimination. As I have argued, we may decide in the end to accept a gap in the law and morality of discrimination, given the administrative and privacy costs of legal intervention, but a general theory would at least identify the factors weighing in favor of legal prohibition. Yet the recent flurry of papers on the subject of customer discrimination, including Bartlett and Gulati, almost entirely avoids the basic theoretical question, why prohibit discrimination? They do so because it seems plausible that whatever makes firm or employer discrimination wrong will make customer or employee discrimination wrong.

By contrast, I worry that we need to start with a general antidiscrimination theory, or a set of them, if we are to successfully address the sub-issue of customer discrimination. A general theory might reveal that the wrong of discrimination is domain specific and that one cannot, without more analysis, analogize between domains. We see the point most directly if we examine in detail the Bartlett and Gulati claim that, despite their general point, some customer discrimination is “acceptable” or even “desirable.”8 They offer these five examples: (1) the Montgomery, Alabama bus boycott of the mid-1950s to protest segregated seating; (2) a 2005 “girlcott” of Abercrombie & Fitch to protest their advertising; (3) a boycott of Russian vodka to protest Russian anti-gay laws; (4) customer support of black-owned businesses by BLACKOUT and FUBU (“For Us By Us”); and (5) federal government minority set-aside requirements for contractors.9 Let us zero in on what could make these exceptional instances of customer discrimination “acceptable” or “desirable.”

I think we can disregard the first two examples as not strictly relevant. In these cases, the boycotters are not acting against the seller because of race or sex (or any other prohibited category). Bus boycotters were not shunning buses for the reason that they were owned or driven by whites, but because they made black passengers give up front seats when white passengers wanted them. Objecting to a demeaning characteristic of the bus service—that it discriminates on the basis of race—is not itself discrimination on the basis of race. Nor was the “girlcott”—they did not care (I assume) if Abercrombie & Fitch or their advertising firm was owned or managed by women or men but objected to the sexist and harmful images in the advertising.

The Russian boycott is potentially different, but only potentially. It would be the same as the “girlcott” if the owners of the boycotted vodka brand played an important role in creating the Russian anti-gay laws, either because the companies acted as private lobbyists for the law or because the company owners are so tied to the Russian government that enacted the law that we cannot distinguish them from government officials. But suppose a Russian vodka manufacturer is distinct from the government and did not push for the law. If so, then the purpose of the consumer boycott in this instance is simply to give the company, which did nothing wrong, an incentive to convince the government to change the law. So the boycotter really is objecting to the vodka not because of what the manufacturer did, but because of what the manufacturer is—i.e., what the owners and operators are—Russian. That is discrimination on the basis of national origin. Even when a secondary boycott like this has a good goal—the repeal of oppressive laws—there is something complex about the utilitarian calculus of applying pressure to someone not at fault for the law’s existence, indeed, who might have opposed it. Especially so when the pressure is brought to bear on the basis of a classification such as national origin.

Bartlett and Gulati assume that international boycotts of goods based on national origin are presumptively wrongful, but that the presumption may be overcome by having the right motive for the boycott (perhaps just when the motive is to fight discrimination). But it seems that a simpler and possibly more compelling answer would involve the underlying rationale for the ban on national origin discrimination. This rationale supplies the presumptive reason to condemn boycotters, but if we understood it completely we might see that it does not apply to this boycott.

For example, perhaps national origin discrimination (and other discrimination) is objectionable only when it contributes to the subordination of a group, keeping it low in the social hierarchy, threatening to turn the group into a caste of low rank. (As we shall see, Bartlett and Gulati briefly endorse something like this view). On this account, the law prohibiting employers from engaging in national origin discrimination is justified because some social groups defined by national origin are subordinated in American society or would be subordinated but for the law. Of course, this rationale would not seem to apply to discrimination against those whose national origin is, say, Scotland, Australia, or Canada. But that reply does not necessarily prove the rationale is wrong because there might be political and administrative advantages to prohibiting all national origin discrimination rather than to define the particular origins at risk of subordination at any given time. We trade off the lack of perfect fit with the advantages of the simpler rule. Every plausible rationale may turn out to be over- and/or under-inclusive in some respect, and anti-subordination theory might do as well at justifying the law as any competing theory.

If so, then the simpler reason for allowing the Russian vodka boycott may be that the national origin theory is a very bad fit for international consumer boycotts and particularly those aimed at Russia. For Americans to participate in such boycotts does not create much risk of subordinating American citizens or residents with a national origin of Russia, both because there is not much risk of subordination of Russian Americans and because an international boycott of Russian-produced vodka is not a good tactic for bringing about such subordination. And though an international boycott of Russian goods harms Russians in Russia, it is a stretch to say that the boycott risks subordinating them in the sense of making them an American or international caste of low rank. So we might think it is not necessary to craft a special justification for the Russian vodka ban. We might just think the deeper theory for prohibiting national origin discrimination does not apply here.

Of course, what is true of Russia might not be true of all nations. There might be some nation in which there is a risk of subordination of American residents from that nation and for which an international consumer boycott against the nation would serve to further such subordination. But we still might not want a law to prohibit any international consumer boycotts. If we care about how well the rule “fits” the rationale, because we trade off the lack of fit with the advantages of the simpler rule, then there is no reason to assume that the trade-off in the domestic arena is the same as the trade-off in the international arena. National origin discrimination in domestic employment and domestic housing might pose a substantial risk of subordination for a substantial number of Americans, whereas national origin discrimination in international consumer boycotts poses only a trivial risk of subordination for only a few Americans. If so, then for this rationale, one does not need to inquire into the merits of the boycott to decide that it is not worth the administrative costs to prohibit national origin discrimination in consumer boycotts of foreign goods.

For my purposes, nothing turns on accepting anti-subordination as the appropriate rationale. Other rationales may produce different results. But that just proves my point: It helps to know what the underlying antidiscrimination theory is before deciding how to treat customer discrimination, including in the boycott of Russian vodka.

Bartlett and Gulati implicitly acknowledge the point in their last two examples of “acceptable” or “desirable” customer discrimination: (4) BLACKOUT and FUBU; and (5) government set-aside programs. BLACKOUT encourages customers to spend on and invest in black-owned businesses, while FUBU is a sportswear brand designed to draw customers who want to patronize black-owned businesspersons.10 Government set-aside programs favor racial minorities in government contracting.11 Bartlett and Gulati recognize these as examples of customer-based race discrimination. We might add that customers sometimes patronize restaurants or bars because the owner or manager shares the same minority ethnicity or nationality. Some go to a restaurant because it is run by a woman chef. Is this kind of discriminatory patronage “acceptable” and, if so, why? Here, and nowhere else in the article, Bartlett and Gulati refer briefly to the anti-subordination rationale of the Equal Protection Clause.12 They say: “A truly ‘color-’ or ‘gender-blind’ approach to combatting customer discrimination would not distinguish these instances from the rest. We would allow these practices, however, because they diminish gender and racial inferiority and subordination rather than contribute to it.”13 Here they cite some of the academic work on the constitution’s Equal Protection guarantee. That work discusses anti-subordination, which is one general theory of antidiscrimination law.

Bartlett and Gulati are surely right that the anti-subordination theory distinguishes consumer behavior favoring African Americans from consumer behavior averse to African Americans. As I said, that proves my point, which is that there is a linkage between the normative status of customer discrimination and the general normative theory of discrimination and we need to explore the latter in order to understand the former. Another possible theory, for example, is that discrimination is wrong when it is based on outgroup animus, but not when based upon a special affinity for the ingroup. I do not advocate such a theory, but one could plausibly decide that negative preferences such as hatred, spite, and sadism are a morally objectionable reason for action, while positive preferences—altruism or love—are not, even when a person exhibits more altruism towards some than others. If this were the right theory, then it would be normatively acceptable in the United States for members of, say, a white ethnic group or a Protestant denomination to patronize establishments that shared their ethnicity or religion. I assume Bartlett and Gulati would reject such a conclusion, as the anti-subordination theory does (because the favoritism of the dominant group toward itself is part of a system of subordination), but again, we know this only because Bartlett and Gulati point to a general normative theory.

Of note here, the Supreme Court has famously moved away from the anti-subordination idea, in favor of an anti-classification rationale for Equal Protection.14 The theory that it is always presumptively wrong to classify according to race makes it far more difficult to distinguish consumer boycotts based on whether they favor or disfavor historically disadvantaged—subordinated—races. Anti-classification theory puts affirmative action in a precarious position, and the government set-aside program Bartlett and Gulati discuss is a form of affirmative action. Given the current Court, the extension of the antidiscrimination regime into customer discrimination might put these same practices into question. Bartlett and Gulati would allow subordinated minorities to build their capital collectively by exclusively patronizing sellers of the same group, which is the same as discriminating against other groups, but if the law were changed to reach customers the Supreme Court would likely find this behavior is just another invidious form of racial discrimination. And thus, again, the way we treat and should treat customer discrimination depends on the fundamental theory for justifying the antidiscrimination norm.

I conclude this Comment with an extended discussion of an alternative example: sex discrimination in dating, sex, and long-term relationships such as marriage. In discussing online dating, Bartlett and Gulati hint at the full implications of their argument when they cite Lawrence v. Texas for the proposition that “[w]ho someone dates is a very personal matter, integral to an individual’s social and sexual identity.”15 They cite Griswold v. Connecticut and its progeny for the claim that “the freedom to make one’s own choices with respect to one’s sexuality . . . is one of this nation’s most protected constitutional rights.”16 And they say this is true “[w]hether or not sexual desire is as inherent to one’s identity as is commonly assumed.”17 The implication of Bartlett and Gulati’s analysis is that, but for efficacy and privacy concerns, which push back against the case for regulation, it would be appropriate to prohibit discrimination in dating and sex, possibly even long-term partnerships.

Much of the discussion so far has focused on race discrimination, but let us now focus on sex discrimination. Without an underlying theory explaining when discrimination is wrong, analogical reasoning suggests that there is a prima facie objection to discriminating in dating, sex, and marriage on the basis of sex. To prefer women as sex partners and therefore partners for dating and marriage is to discriminate against men. To prefer men is to discriminate against women. In each case, it does not matter if the discriminator is a man or woman, a hetero- or homosexual. Even for those who engage in sexual relations with members of both sexes, the implication is that they are guilty of wrongful discrimination if they seek out or accept offers from one sex more than the other, acting on a preference for men or for women. On this logic, the only morally acceptable, non-bigoted form of sexual interest is unbiased bisexuality.18 Presumably the same analogy could be applied to age discrimination.19

Of course Bartlett and Gulati leave plenty of room for justifying the legality of sex and age discrimination in dating, sex, and marriage. In their view, it is permissible to leave discrimination unregulated when there are strong countervailing values at stake, such as efficacy and privacy costs. But this analysis strikes me as flawed. I believe that racial discrimination in dating, sex, and marriage is wrong, but I doubt that there is anything even presumptively wrong about sex and age discrimination in dating, sex, and marriage. I say this tentatively, but what I need to persuade me otherwise is an underlying theory about the wrong or harm of discrimination that applies just as strongly in this context as any other. Maybe race, sex, and age discrimination are highly analogous to each other in some contexts but not in others. We can only know by resorting to a general normative theory of discrimination.

Perhaps my skepticism will make more sense if I illustrate it with such a theory. I once offered to explain race discrimination as a pathological form of group status competition.20 My claim is that individuals in racial groups compete for status by trying to raise the relative rank of their group. Discrimination is an indirect means of raising one’s group status by lowering the status of others. The point is not generally to avoid association with members of other races, as Gary Becker once suggested,21 but to avoid particular associations—those where one occupies what is perceived as an equal or lower rank—and to seek out associations where one occupies what is perceived as a higher rank. Employment discrimination contributes to lowering the status of the victims because it: (1) is directly demeaning; and (2) inflicts an economic loss in a culture that respects wealth. Romantic and sexual discriminations work in the same way, most obviously by demeaning those deemed as categorically unfit partners and indirectly by preventing marriages that would spread family wealth and undermine racial categories. Moving from the descriptive to the normative, I claim that this status competition is zero sum, meaning there is no social benefit to moving relative status around in this way, so that the resources invested in discrimination (forgoing otherwise productive exchanges) are wasted. Thus, the theory justifies laws prohibiting race discrimination.

The status-production theory is only one of many explanations, but let us assume for now it is correct. Let us also assume that status production explains sex and sexual orientation employment discrimination in the same way (untested hypotheses), and thereby justifies a ban on discrimination by sellers and employers. Notwithstanding these assumptions, the theory might not condemn all types of customer discrimination. In particular, the status theory might allow sex discrimination in the selection of sex partners.

For example, one might say that sex discrimination specifically about sex is not the result of status competition. Although I am assuming here that sex discrimination in employment fits this model,22 it is not at all clear that people exhibit biased sexual orientations (“biased” meaning orientations biased toward a particular sex) with the motive or effect of improving the status of their sexual group or the sexually desired group. Plausibly, gay men do not generally avoid women as sex partners in order to produce status for gay men at the expense of everyone else or status for men at the expense of women. Plausibly, heterosexual men do not generally exclude men as sex partners in order to produce status for straight men at the expense of everyone else or status for women at the expense of men. These claims are consistent with the view that some heterosexuals discriminate in other respects, say employment or housing, against gays and lesbians in order to produce status for heterosexuality, or that heterosexual men discriminate against gay men in employment or housing in order to enforce gender roles, which operates also to subordinate women. My point is simply that people may have genuinely sexual preferences—not only status preferences—that reflect the greater sexual enjoyment or pleasure they receive from having a sex partner of a particular sex.

Nor does exclusion from dating, sex, and marriage on account of sex have the same subordinating effects. There is plausibly no class dimension to sexual orientation (unlike race), so that the refusal of a heterosexual to engage in same-sex dating that might lead to long-term partnership does not serve to impoverish gays and lesbians. Also, it is not clear that sex discrimination in dating, sex, and marriage is inherently demeaning to those excluded; unlike race, exclusion might convey no inherent message of inferiority. Why not? There is much one could say about how to interpret an action to determine whether it demeans another.23 But if one cares about consequences, then the meaning that matters is audience meaning, the one most people actually receive.24 As matters stand, while being undesired is disappointing, I do not believe that people typically perceive an insult—that they are being demeaned—at being categorically excluded from sexual relations because of the status of being male or female (as opposed to being excluded because of race). Nearly everyone accepts that there are many reasons for short- and long-term incompatibility of sexual and romantic partners other than inferiority, and whether a person is male or female is generally understood as a fact relevant to compatibility and not as expressing the inferiority of the undesired sex. (To be clear, by contrast, expressing a categorical preference for sex with a cisgender partner is understood by some to express the inferiority of transgender individuals, though the issue is contested).25

Again, my application of the status production theory is tentative and may be wrong. The important point, however, is that one needs to explore this and other general theories of discrimination before concluding, merely by analogy, that discrimination in one role is exactly the same as discrimination in another.


*   *   *   *   *


Bartlett and Gulati should be commended for shining a spotlight on the fact that antidiscrimination law contains a large omission that is not obviously justified—the legality of customer discrimination. They do not claim to offer a definitive account of the subject nor to resolve all the issues it raises. But even if they intend only to start a conversation, I believe an essential part of the dialogue is to ask the deep and general question of why law regulates discrimination. Only a general theory can fully illuminate what types and domains of discrimination should be prohibited.

  1. [1]. See generally Katharine T. Bartlett & Mitu Gulati, Discrimination by Customers, 102 Iowa L. Rev. 223 (2016). Bartlett and Gulati propose that certain firms should have “a duty ... to refrain from facilitating harmful discrimination by their customers and, in some cases, to structure customer choices in order to diminish the effects of their discriminatory biases.” Id. at 228.

  2. [2]. There was a thin literature on the problem before the Bartlett and Gulati article. See, e.g., Tarunabh Khaitan, A Theory of Discrimination Law 195–214 (2015). See generally Michael Blake, The Discriminating Shopper, 43 San Diego L. Rev. 1017 (2006); Elizabeth F. Emens, Intimate Discrimination: The State’s Role in the Accidents of Sex and Love, 122 Harv. L. Rev. 1307 (2009); Kimani Paul-Emile, Patients’ Racial Preferences and the Medical Culture of Accommodation, 60 UCLA L. Rev. 462 (2012). Since their article, there seems to be increasing interest. See generally Lee Anne Fennell, Searching for Fair Housing, 97 B.U. L. Rev. 349 (2017); Saul Levmore, Title VII to Tinder: Law’s Antidiscrimination Asymmetry and Occasional Market Superiority, 68 Ala. L. Rev. 877 (2017); Heather M. Whitney, The Autonomy Defense of Private Discrimination (Feb. 22, 2017) (unpublished manuscript),

  3. [3]. See George S. Geis, Gift Promises and the Edge of Contract Law, 2014 U. Ill. L. Rev. 663, 667 (noting that, although there are many exceptions to the consideration doctrine that “formally bars gift promises from the domain of contract law,” “contract law has refused to convert all gift promises into binding obligations”); James Gordley, Enforcing Promises, 83 Calif. L. Rev. 547, 569–70 (1995) (noting that “courts do not enforce all promises, or even all those that further an economic activity”).

  4. [4]. See, e.g., Jed Rubenfeld, The Riddle of Rape-by-Deception and the Myth of Sexual Autonomy, 122 Yale L.J. 1372, 1396 (2013) (noting that rape was traditionally defined to require force and many courts adhere to the rule that “fraud is not force” (quoting State v. Brooks, 76 N.C. 1, 3 (1877))). More examples could be added. Plagiarism, line breaking, and hate speech all sometimes transgress common morality without creating legal liability.

  5. [5]. Bartlett & Gulati, supra note 1, at 241.

  6. [6]. See Levmore, supra note 2, at 893–99.

  7. [7]. Deborah Hellman, When is Discrimination Wrong? 90 (2008) (emphasis added). The normative literature is vast. See generally, e.g., Philosophical Foundations of Discrimination Law (Deborah Hellman & Sophia Moreau eds., 2013); Larry Alexander, Book Review, 125 Ethics 872 (2015) (reviewing id.).

  8. [8]. Bartlett & Gulati, supra note 1, at 242.

  9. [9]. Id.

  10. [10]. Id.

  11. [11]. Id.

  12. [12]. Id.

  13. [13]. Id.

  14. [14]. See Bradley A. Areheart, The Anticlassification Turn in Employment Discrimination Law, 63 Ala. L. Rev. 955, 988–95 (2012) (discussing Ricci v. DeStefano, 557 U.S. 557 (2009)). See generally Reva B. Siegel, Equality Talk: Antisubordination and Anticlassification Values in Constitutional Struggles over Brown, 117 Harv. L. Rev. 1470 (2004) (discussing Brown v. Bd. of Educ., 347 U.S. 483 (1954)).

  15. [15]. Bartlett & Gulati, supra note 1, at 243 & nn.82–83.

  16. [16]. Id. at 243 & n.84.

  17. [17]. Id.

  18. [18]. Some might want to rely here on a distinction between the personal and the economic. That is not available to Bartlett and Gulati because they are rejecting such a distinction. But note how the same issue arises even if the distinction were sensible. We need only imagine a world (such as parts of Nevada) where prostitution is legal. The question then arises: can a prostitute refuse sexual business based on the sex of the customer? Does every sex worker have to sign on for perfectly unbiased bisexuality?

  19. [19]. If the antidiscrimination norm applies to individuals in all social roles, and age is a prohibited category, the implication is that adults should not discriminate in the selection of other adults for dating, sex, or long-term partnership on the basis of age. One might reply that the Age Discrimination in Employment Act (“ADEA”) merely prohibits discrimination against those over age 40, which might imply the permissibility of other age discrimination. But the law still implies that people in their twenties are ageist and wrong to pursue a preference for people under 40 in any aspect of their personal life. One might also reply that young people seeking life partners should be allowed to discriminate on the basis of infertility or life expectancy, which are correlated with age. But this generalization is exactly the kind of age profiling the ADEA seeks to end among employers. Just as some older workers may be highly productive, some older potential partners may be fertile and have a long life expectancy. The analogy implies that age may not be used as a proxy.

  20. [20]. See generally Richard H. McAdams, Cooperation and Conflict: The Economics of Group Status Production and Race Discrimination, 108 Harv. L. Rev. 1003 (1995).

  21. [21]. See generally Gary S. Becker, The Economics of Discrimination (2d ed. 1971).

  22. [22]. To read my argument for this in an unposted working paper, see Richard H. McAdams, Economic Theories of Discrimination: The Positive and the Normative (2017) (unpublished manuscript) (on file with author).

  23. [23]. See generally Hellman, supra note 7 (discussing discrimination).

  24. [24]. See Richard H. McAdams, The Expressive Powers of Law 19–21, 243–48 (2015).

  25. [25]. I will not even sketch the status analysis for age discrimination in dating, sex, and long-term partnerships. Many arguments present themselves; some point in favor and some against the analogy to race.


Bernard D. Meltzer Professor of Law, University of Chicago Law School.

For discussions and comments on earlier drafts, I thank Kenworthey Bilz, Jessica Clarke, Lee Anne Fennell, Andrew Leipold, Saul Levmore, Arden Rowell, and Heather Whitney.